Minister for Foreign Affairs Dr Vivian Balakrishnan's Speech at AI Engineer Singapore, 16 May 2026

1. Home
2. Newsroom
3. Press Statements, Transcripts & Photos
4. Minister for Foreign Affairs Dr Vivian Balakrishnan's Speech at AI Engineer Singapore, 16 May 2026

Transcripts

16 May 2026

Minister for Foreign Affairs Dr Vivian Balakrishnan's Speech at AI Engineer Singapore, 16 May 2026

Hi, good morning everyone.

I feel like an impostor here. For those of you who do not know me, I am a retired eye surgeon. But I took a detour into politics, for perhaps too long. But I have always retained an interest in getting things done, building things, fixing things.

Since I do not get to operate on eyes anymore, I assemble watches, I reprogram appliances, and now there is some other stuff which is what I am going to talk about today. I wanted to explain why I did it, and the implications of this. I think with this audience, you will get it straightaway.

But, let me jump to the end. These are the three key messages, which - if you forget everything else I have said, just bear these things in mind.

We are now at an age where you can outsource a lot of stuff. Calculations, computation, memory, replication, dissemination of knowledge. The one thing which you cannot outsource is your personal understanding. If you are in a position of authority, you can delegate work - you cannot delegate accountability. So, remember the personal element - understanding and accountability.

The next point - and I would refer you to a nice short letter published in the Financial Times by Professor Neil Lawrence, DeepMind Professor of Machine Learning, University of Cambridge. He is the Professor of Machine Learning. And you know, there is a lot of hype about AI models, data centres, top-down systems, rules, governments. That is macro. But his hypothesis is that real value for the economy and society is created at the ground level. Workflow by workflow, sector by sector, department by department, and in fact, at the individual level.

What this means is that the real payoff is when ordinary people - teachers, lawyers, technicians, managers, doctors, even ministers - are actually using the tools which are already available, already invented. People who know their jobs and are empowered by these tools. That is how you create real value for society and for the economy.

So I am looking at decentralisation, individualisation and bespoke models. I am talking about making yourself better at what your day job is, and even better still - re-engineering the workflows of your life. That is where the real value boost is.

And the third takeaway is that I sincerely believe the barriers for achieving all these have collapsed. The tools have already been invented. It is a matter of getting people to understand what tools are out there, assemble their own tools, and put themselves on a completely different trajectory.

So now let us do the fun part of my adventures.

Now, my personal agent first came to life almost exactly three months ago. Given my job, I knew that OpenClaw was not practical because security was an issue. Then, someone else pointed to NanoClaw. I think we are going to hear from NanoClaw Creator Gavriel Cohen after this.

As a geek and as a tinkerer myself, I like stuff which I can grasp. So the fact that NanoClaw has a very short code base, which even an idiot like me can read and sort of understand; the fact that it is containerised, and as a surgeon I know that there is no such thing as a routine operation, and things will go wrong, things will break - and when they do break, hopefully you want them to break within barriers. So, the containerisation and the understandability parts were vital for me.

Anyway, simple - got on GitHub, downloaded this stuff. And the other attractive part about it is that there are no configurations. In fact, you rely on an Large Language Model (LLM) to do all the bespoke, tailored customisations, you realise everyone running an instance of NanoClaw is running an individualised system. Now, that is both good and also has its share of complications.

But anyway, let me tell you what I did with it.

So NanoClaw provides the platform. It allows me to communicate through WhatsApp with my agent; that part is not rocket science. The thing which I was really after was: how could I use it for my daily life?

Let me give you an idea of my daily life. This month, I am visiting 12 countries. I will therefore have to meet hundreds of people. I will have to understand the country's economy, geography, culture, history, war and peace. I need to know people as individuals and not just something from a brief. There is a huge cognitive overload on every single diplomat. The question is, how can I turbocharge this process so that if I need a fact or a factoid, I can get it, I can get it anywhere, and I can go down a rabbit hole, if need be.

The LLMs are useful for analysis, for abstraction, for expression and certainly for drafting briefs, drafting speeches, formulating answers to questions including, I must add, parliamentary questions. And three months ago, it was extremely impressive to see both the questions and answers it generated during debates in Parliament three months ago.

So, it communicates with me through WhatsApp, through this software called Baileys. I suspect it is not entirely keeping with what Meta or WhatsApp would like us to do, because it is actually simulating the way we get WhatsApp to work in our browsers or in our laptops - so it is a pseudo-terminal, in a sense.

Then, the bit which I believe is the real frontier for people like me is memory. And fortunately, I came across this piece of software called mnemon. I still have not met the developer, so I do not really know, but a memory system with graphs. So, it has got entities: causality, temporal relationships, and semantics. And because I did not want to be confined to just keyword searches, the fact that I could run Ollama locally with an embedding model means I also have semantic search built in. So, with these elements - whisper.cpp is the part that is easy, because with WhatsApp, I did not want to only have to type, I wanted to be able to speak and it can respond to me. And of course, my dream is one day to just have my agent answer supplementary questions in Parliament. I am not sure about the legality of that, but if it happens you know that I shared the idea with you first.

But the point is I was now able to curate material, speeches, transcripts particularly of my own contributions, get it into the system, digest it, extract it, put into that memory database.

And then around the same time, Andrej Karpathy came up with his LLM-supervised Wiki generation, so I added that in as well.

Then for the user experience, the user interface, I used Obsidian, partly because Obsidian allows me to use the Apple iCloud, and that immediately means I have got a personal cloud, and all the Wikis which are extracted from this personally curated database become available to me. Because remember, I started off by saying: the key is personal understanding.

So, I have got a memory system, I have got a communication system, I have got an analysis system. All nice in theory.

But what I am here to share with you is that in the last three months, I have found it incredibly useful - meeting people, travelling, first cut of a speech, even today's presentation were generated by Claude. It has turbocharged the pace at which things can be done.

As a practitioner - not as an engineer, but as a practitioner with a day job, it is useful. And I can attest to its usefulness because I can honestly tell you, I have not dared to switch it off. NanoClaw, unfortunately, has moved from Version 1 to Version 2. When Version 2 came on, because the transition is not at all smooth, I left Version 1 working and I put Version 2 on another computer. And I should also add: all this stuff - my most daily-used agent - is running off a Raspberry Pi, which is at least two or three years old. All it has is eight gigabytes of RAM. You see my point about accessibility, personalisation, relevance and use.

This is my point: the barriers have fallen. Because I did this without writing Claude, Baileys, mnemon, whisper.cpp, or the credentialling system. You know, there is this whole thing about vibe-coding. I would not even dare to claim I was vibe-coding. I was just assembling tools, just tool assembly. Yes, I have gone through the code - you know, NanoClaw insists that you approve when every time you give dash access to the agent, so I do scan through it. It does help if you understand coding.

In a sense, my approach to all this has been to learn by doing. It is not enough to sit down and read, get the headlines, get the summaries done. If you are interested in anything, get your hands dirty. You learn best by doing. And because the barriers to entry have come down so dramatically, everyone should embark on their personal experiments. And Claude came up with this quote,: you cannot govern a technology that you have only been briefed on. You had better get your hands dirty, and then you understand both the potential and the limits, and the problems.

A few other digressions down here: there are constraints. So, for instance, depending on LLMs - and quite frankly, the prices for which the AI majors are currently charging us, I think we all know we are enjoying in effect a subsidy. Tokens are not cheap, compute power is limited, electricity prices have risen, wars do not help. And we should beware of just trying to throw every problem and every step in a solution at an LLM. It reminds me of the old proverb: for a man with a hammer, everything looks like a nail. In fact, there are both good economic and design advantages so that you use LLMs, but do not forget there is still a role for deterministic systems. There is still a role for expert rule-based systems. My personal belief as a biologist, is for in the end, some kind of neurosymbolic system, rather than just the LLM model. And I have some sympathy for Yann LeCun, who says that LLMs are great, but actually that is not the way we have solved it in nature. If you look at the human brain, actually I suspect we have less layers of computation in the human brain than in many of the LLMs that we have today.

And I can tell you as an eye surgeon, the cortical computation for vision, for language, for cognition, are often based on far more efficient structures than the energy-gobbling systems which we have today. The point I am making, and where I am agreeing with Yann LeCun, is that in the end, these are pattern recognition systems with attention, with memory. Out of what looks like simple fundamental abilities is emergent behaviour, which gives you conceptual understanding, which gives you language, which gives you the ability to do things. So, my point is: this is a field which is still exploding. Therefore, approach this with humility, approach this by just doing your best, improving the productivity of your daily job - but understand that we are perhaps one of the most privileged generations to be living through a revolution.

Tools matter more than models.

And finally, memory. It is very human, and I think it is the great unsolved part of this frontier.

On security, I am not going to belabour this. Just as an aside - even if you hack my system, the most you will get is my phone number. You will get summaries of foreign policy which I have curated anyway.

Now, that is one way of addressing security - by making sure you only put what is already open-source, what is already published and you subject your systems to a level of transparency and scrutiny that can be withstood.

But do not forget security remains paramount. In fact, the complication to the dissemination of AI is going to be commercial competition, national security, cybersecurity, and superpower contestation. These are the political factors that are going to affect the availability, the speed, and the dissemination of AI of the future. This, again, is a separate political talk, well worth a deep dive.

I am a believer in deployment at the edge. I am a surgeon. I believe in doing. I believe in fixing. I believe that is where lives are saved and value is created.

Therefore the public policy goal is the democratisation of these tools. That is why you will see in the Economic Strategy Review Committee, Deputy Prime Minister Gan said that Singapore is not likely to be at the frontier of model development. But we can be at the frontier of deployment at scale. So, democratisation.

Therefore, if that is what we believe, then it must be a decentralised, ground-up approach. And that is why I am here today, because I found out this conference was organised less than three months ago: 65Labs, all the people you meet here - this is not even their day job. It is a hack. But this is the way I believe the future is going to be created.

So, thank you all for being here, thank you for being part of this journey. Have a wonderful day, a wonderful future.

Thank you very much.

. . . . .

MINISTRY OF FOREIGN AFFAIRS

SINGAPORE

16 MAY 2026